Fig. 3.

Step by step CMR-ECGI workflow. a The participant lies in the supine position for a 5-min recording at rest and an inflatable gilet is worn over the electrode vest to ensure good skin–electrode contact. b 256 ECG signals (torso potentials ϕ_T) are collected from the dry textile electrodes and amplified using the g.HIamp at a sampling frequency of 2400 Hz and processed using g.Recorder. c CMR scanning is performed at 3T to include T1 mapping, T2 mapping, LGE and ECV and transaxial HASTE used for cardiac segmentation and landmark localisation. d ECG signals are post processed using in-house Matlab software. e Heart-torso geometry (A) is generated from the HASTE via the reconstruction of individual cardiac meshes and imputation of the electrode markers using Amira software. f Heart-torso geometry (A) is combined with the torso potentials (ϕ_T) to solve the ‘inverse solution’ of ECGI (ϕ_T = Aϕ_E) to reconstruct epicardial potentials (ϕ_E) at 1000 individual cardiac sites which are used to compute AT, RT and ARI per site and generate isochrone maps of cardiac excitation. In this example of a 75-year-old participant the earliest site of activation is in the posterolateral lateral wall of the LV. Ao aorta, ARI activation recovery interval, ECV extracellular volume, F frequency, LA left atria, LGE late gadolinium enhancement, LV left ventricle, mV millivolts, ms milliseconds, PA pulmonary artery, RA right ventricle, RV right ventricle, UEG unipolar electrogram. Other abbreviations as in Fig. 1